RWD Analysis Using Open Source Software

Introduction to OMOP CDM and R Packages for Analysis

PHUSE Japan Open-source Technology Working Group

December 5, 2025

Slides Available Here

English: https://nissinbo.github.io/phuse-sde-tokyo-2025-ost-omop/en

Japanese: https://nissinbo.github.io/phuse-sde-tokyo-2025-ost-omop/ja

Introduction to OMOP CDM

What is OMOP CDM?

Observational Medical Outcomes Partnership Common Data Model

  • A standardized data model for unified analysis across diverse RWD sources
  • Enhances reproducibility through common schema and standardized vocabularies
  • Developed and maintained by the OHDSI community

OHDSI

Note

In 2025, a PHUSE Working Group was established to promote OMOP adoption in the pharmaceutical industry

Key Features of OMOP CDM

  • Advantages: Facilitates comparative studies across different data sources
    • Unified terminology through standardized vocabularies
    • Designed with minimal tables necessary for observational research
  • Challenges: Complexity of transformation process
    • Requires ETL processes specific to each data source
    • Difficulty in mapping to standard terminologies

OMOP CDM Structure (v5.4)

  • Clinical data: Person, Observation Period, Visit Occurrence, …
  • Health system: Location, Care Site, Provider
  • Vocabularies: Concept, Vocabulary, Concept Relationship, …

OMOP CDM

Core Table: Person

Basic patient information and demographics

Field Description
person_id Patient ID
gender_concept_id Gender
year_of_birth Birth year
race_concept_id Race
ethnicity_concept_id Ethnicity

Tip

All clinical events are linked through person_id

Core Table: Visit Occurrence

Healthcare facility visit and admission information

Field Description
visit_occurrence_id Visit identifier
person_id Patient ID
visit_concept_id Visit type (inpatient/outpatient/emergency)
visit_start_date Visit start date
visit_end_date Visit end date

Core Table: Condition Occurrence

Disease and symptom diagnosis information

Field Description
condition_occurrence_id Diagnosis identifier
person_id Patient ID
condition_concept_id Standard concept ID for condition
condition_start_date Diagnosis start date
condition_type_concept_id Record source (EHR/claims)

Core Table: Drug Exposure

Drug exposure information

Field Description
drug_exposure_id Drug exposure identifier
person_id Patient ID
drug_concept_id Standard concept ID for drug
drug_exposure_start_date Exposure start date
drug_exposure_end_date Exposure end date

Code Mapping

In OMOP, various codes are often mapped to standard concepts (not mandatory)

  • Standard concepts: Defined by SNOMED CT , RxNorm, etc.
  • Non-standard concepts: Codes in source data such as ICD10, LOINC
  • Concepts can be searched using ATHENA web tool

Example: Hypertension

  • Standard: SNOMED 38341003
  • Non-standard: ICD10 I10, MeSH D006973

Analyzing OMOP with R

What is HADES?

Health Analytics Data-to-Evidence Suite

  • A collection of R packages specialized for OMOP CDM data analysis
  • High interoperability (works seamlessly when using HADES packages together)
  • Actively developed by two organizations: OHDSI and DARWIN EU

HADES OHDSI DARWIN EU

HADES Package List

HADES Packages

As of December 2025, 41 packages are registered in HADES!

Example Workflow Using HADES

End-to-end analysis workflow is achievable

%%{init: {'theme':'base', 'themeVariables': {'fontSize':'30px'}}}%%
graph LR
    subgraph DataQuality["Data Quality"]
    B1[DataQualityDashboard]
    B2[Achilles]
    end
    
    subgraph CohortDefinition["Cohort Definition"]
    C1[Capr]
    C2[CohortGenerator]
    end
    
    subgraph CohortDiagnostics["Cohort Diagnostics"]
    D1[CohortDiagnostics]
    end
    
    subgraph PatientCharacteristics["Patient Characteristics"]
    E1[FeatureExtraction]
    E2[Characterization]
    end
    
    subgraph Estimation["Estimation"]
    F1[CohortMethod]
    F2[EvidenceSynthesis]
    end
    
    A[OMOP Database] --> B1 & B2
    A --> C1 & C2
    C1 & C2 --> D1
    D1 --> E1 & E2
    E1 --> F1 & F2
    
    style DataQuality fill:#FCE4EC,stroke:#C2185B,color:#000
    style CohortDefinition fill:#E8F5E9,stroke:#388E3C,color:#000
    style CohortDiagnostics fill:#FFF3E0,stroke:#F57C00,color:#000
    style PatientCharacteristics fill:#F3E5F5,stroke:#7B1FA2,color:#000
    style Estimation fill:#FFF9C4,stroke:#F9A825,color:#000
    
    style A fill:#E3F2FD,stroke:#1976D2,color:#000
    style B1 fill:#FFEBEE,stroke:#C62828,color:#000
    style B2 fill:#FFEBEE,stroke:#C62828,color:#000
    style C1 fill:#C8E6C9,stroke:#2E7D32,color:#000
    style C2 fill:#C8E6C9,stroke:#2E7D32,color:#000
    style D1 fill:#FFE0B2,stroke:#EF6C00,color:#000
    style E1 fill:#E1BEE7,stroke:#6A1B9A,color:#000
    style E2 fill:#E1BEE7,stroke:#6A1B9A,color:#000
    style F1 fill:#FFF59D,stroke:#F57F17,color:#000
    style F2 fill:#FFF59D,stroke:#F57F17,color:#000

Let’s Try It Out! 😀

Setup

Install R packages

install.packages(c("duckdb", "here", "CDMConnector", "OmopSketch", 
                   "PatientProfiles", "IncidencePrevalence", "CohortSurvival"))

Download sample data

library(CDMConnector)

Sys.setenv("EUNOMIA_DATA_FOLDER" = here::here())
downloadEunomiaData("GiBleed")

CDMConnector + Basic Operations

CDMConnector

CDMConnector

Database connection and data access

library(CDMConnector)
library(tidyverse)
library(dbplyr)

# Connect to database
con <- DBI::dbConnect(duckdb::duckdb(), eunomiaDir("GiBleed"))

# List tables
DBI::dbListTables(con)
 [1] "care_site"             "cdm_source"            "concept"              
 [4] "concept_ancestor"      "concept_class"         "concept_relationship" 
 [7] "concept_synonym"       "condition_era"         "condition_occurrence" 
[10] "cost"                  "death"                 "device_exposure"      
[13] "domain"                "dose_era"              "drug_era"             
[16] "drug_exposure"         "drug_strength"         "fact_relationship"    
[19] "location"              "measurement"           "metadata"             
[22] "note"                  "note_nlp"              "observation"          
[25] "observation_period"    "payer_plan_period"     "person"               
[28] "procedure_occurrence"  "provider"              "relationship"         
[31] "source_to_concept_map" "specimen"              "visit_detail"         
[34] "visit_occurrence"      "vocabulary"

CDMConnector

Create CDM object

Use cdmFromCon() to create an OMOP-specific object format

cdm <- cdmFromCon(con, cdmSchema = "main", writeSchema = "main")
cdm

cdm Object

Access each table using $

cdm$person |> 
  collect() |> 
  glimpse()
Rows: 2,694
Columns: 18
$ person_id                   <int> 6, 123, 129, 16, 65, 74, 42, 187, 18, 111,…
$ gender_concept_id           <int> 8532, 8507, 8507, 8532, 8532, 8532, 8532, …
$ year_of_birth               <int> 1963, 1950, 1974, 1971, 1967, 1972, 1909, …
$ month_of_birth              <int> 12, 4, 10, 10, 3, 1, 11, 7, 11, 5, 8, 3, 3…
$ day_of_birth                <int> 31, 12, 7, 13, 31, 5, 2, 23, 17, 2, 19, 13…
$ birth_datetime              <dttm> 1963-12-31, 1950-04-12, 1974-10-07, 1971-…
$ race_concept_id             <int> 8516, 8527, 8527, 8527, 8516, 8527, 8527, …
$ ethnicity_concept_id        <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
$ location_id                 <int> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
$ provider_id                 <int> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
$ care_site_id                <int> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA…
$ person_source_value         <chr> "001f4a87-70d0-435c-a4b9-1425f6928d33", "0…
$ gender_source_value         <chr> "F", "M", "M", "F", "F", "F", "F", "M", "F…
$ gender_source_concept_id    <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
$ race_source_value           <chr> "black", "white", "white", "white", "black…
$ race_source_concept_id      <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …
$ ethnicity_source_value      <chr> "west_indian", "italian", "polish", "ameri…
$ ethnicity_source_concept_id <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, …

Basic Data Operations

Distribution of conditions among males born after 1975

cdm$person |>
  filter(year_of_birth >= 1975, gender_source_value == "M") |> 
  left_join(cdm$condition_occurrence, by = "person_id") |>
  summarise(n = n(), .by = condition_concept_id) |>
  left_join(cdm$concept |> select(concept_id, concept_name), by = c("condition_concept_id" = "concept_id")) |>
  collect() |> 
  arrange(desc(n))
# A tibble: 62 × 3
  condition_concept_id     n concept_name           
                 <int> <dbl> <chr>                  
1             40481087   744 Viral sinusitis        
2              4112343   419 Acute viral pharyngitis
3               260139   354 Acute bronchitis       
4               372328   210 Otitis media           
# ℹ 58 more rows

Visualization Too!

cdm$person |> 
  summarize(n = n(), .by = c(year_of_birth, gender_concept_id)) |> 
  mutate(sex = case_when(
    gender_concept_id == 8532 ~ "Female",
    gender_concept_id == 8507 ~ "Male"
  )) |> 
  collect() |> 
  ggplot(aes(y = n, x = year_of_birth, fill = sex)) +
  geom_col(position = "dodge")

Note

tidyverse style data handling is possible!

OmopSketch
Understanding Database Overview

OmopSketch

OmopSketch

Get an overview of the entire database (tibble)

library(OmopSketch)

cdm |> 
  summariseOmopSnapshot() |> 
  tableOmopSnapshot(type = "tibble")
# A tibble: 13 × 3
   Variable           Estimate                [header_name]Database name\n[hea…¹
   <chr>              <chr>                   <chr>                             
 1 General            Snapshot date           2025-11-13                        
 2 General            Person count            2,694                             
 3 General            Vocabulary version      v5.0 18-JAN-19                    
 4 Observation period N                       5,343                             
 5 Observation period Start date              1908-09-22                        
 6 Observation period End date                2019-07-03                        
 7 Cdm                Source name             Synthea synthetic health database 
 8 Cdm                Version                 v5.3.1                            
 9 Cdm                Holder name             OHDSI Community                   
10 Cdm                Release date            2019-05-25                        
11 Cdm                Description             SyntheaTM is a Synthetic Patient …
12 Cdm                Documentation reference https://synthetichealth.github.io…
13 Cdm                Source type             duckdb                            
# ℹ abbreviated name: ¹​`[header_name]Database name\n[header_level]Synthea`

OmopSketch

Get an overview of condition_occurrence (flextable)

cdm |> 
  summariseClinicalRecords("condition_occurrence") |>
  tableClinicalRecords(type = "flextable")

Variable name

Variable level

Estimate name

Database name

Synthea

condition_occurrence

Number records

-

N

65,332

Number subjects

-

N (%)

2,694 (100.00%)

Records per person

-

Mean (SD)

24.25 (7.41)

Median [Q25 - Q75]

23 [19 - 29]

Range [min to max]

[5 to 65]

In observation

No

N (%)

450 (0.69%)

Yes

N (%)

64,882 (99.31%)

Domain

Condition

N (%)

65,332 (100.00%)

Source vocabulary

Icd10cm

N (%)

479 (0.73%)

No matching concept

N (%)

27 (0.04%)

Snomed

N (%)

64,826 (99.23%)

Standard concept

S

N (%)

65,332 (100.00%)

Type concept id

Ehr encounter diagnosis

N (%)

65,332 (100.00%)

OmopSketch

Get an overview of drug_exposure (flextable)

cdm |> 
  summariseClinicalRecords("drug_exposure") |>
  tableClinicalRecords(type = "flextable")

Variable name

Variable level

Estimate name

Database name

Synthea

drug_exposure

Number records

-

N

67,713

Number subjects

-

N (%)

2,694 (100.00%)

Records per person

-

Mean (SD)

25.13 (5.25)

Median [Q25 - Q75]

25 [22 - 28]

Range [min to max]

[7 to 54]

In observation

No

N (%)

251 (0.37%)

Yes

N (%)

67,462 (99.63%)

Domain

Drug

N (%)

67,713 (100.00%)

Source vocabulary

Cvx

N (%)

25,713 (37.97%)

Ndc

N (%)

2,694 (3.98%)

No matching concept

N (%)

35 (0.05%)

Rxnorm

N (%)

39,271 (58.00%)

Standard concept

S

N (%)

67,713 (100.00%)

Type concept id

Dispensed in outpatient office

N (%)

25,713 (37.97%)

Prescription written

N (%)

42,000 (62.03%)

PatientProfiles
Adding Patient Characteristics

PatientProfiles

PatientProfiles

Define a cohort of “patients with bronchitis”

cdm <- cdm |> 
  generateConceptCohortSet(
  name = "bronchitis",
  conceptSet = list("any_bronchitis" = c(260139, 258780)), 
  limit = "all", 
  end = 0
)

cdm$bronchitis |> 
  collect()
# A tibble: 8,232 × 4
  cohort_definition_id subject_id cohort_start_date cohort_end_date
                 <int>      <int> <date>            <date>         
1                    1        245 1991-04-10        1991-04-10     
2                    1        411 1973-11-17        1973-11-17     
3                    1        411 2005-10-13        2005-10-13     
4                    1        782 1921-11-16        1921-11-16     
# ℹ 8,228 more rows

PatientProfiles

Add patient characteristics to cohort

library(PatientProfiles)

# Date of birth, age, sex
cdm$bronchitis |> 
  addDateOfBirth() |> 
  addSex() |> 
  addAge()

# Prior/future observation periods from index date
cdm$bronchitis |> 
  addPriorObservation() |> 
  addFutureObservation()

Note

Nothing much to say here. It’s incredibly simple!

IncidencePrevalence
Calculating Prevalence and Incidence

IncidencePrevalence

IncidencePrevalence

Create “denominator” cohort

library(IncidencePrevalence)

cdm <- cdm |> 
  generateDenominatorCohortSet(
    "denom", 
    cohortDateRange = c(as.Date("2005-01-01"), as.Date(NA))
  )

IncidencePrevalence

Calculate prevalence

cdm |> 
  estimatePeriodPrevalence(
    denominatorTable = "denom", 
    outcomeTable = "bronchitis"
  ) |> 
  plotPrevalence()

IncidencePrevalence

Calculate incidence

cdm |> 
  estimateIncidence(
    denominatorTable = "denom", 
    outcomeTable = "bronchitis"
  ) |> 
  plotIncidence()

CohortSurvival
Survival Analysis

CohortSurvival

CohortSurvival

library(CohortSurvival)

# Sample data for survival analysis
cdm <- mockMGUS2cdm()

cdm |> 
  estimateSingleEventSurvival(
  targetCohortTable = "mgus_diagnosis",
  outcomeCohortTable = "death_cohort"
) |> 
  plotSurvival()

CohortSurvival

cdm |> 
  estimateSingleEventSurvival(
    targetCohortTable = "mgus_diagnosis",
    outcomeCohortTable = "death_cohort",
    strata = list(c("age_group"))
) |> 
  plotSurvival(colour = "age_group")

Summary

  • OMOP CDM
    • Common data model for observational and RWD research
    • Standardizes different databases to enable reproducible analysis
  • R Packages for Analysis
    • Ecosystem centered around HADES
    • Many convenient packages specialized for OMOP analysis

Learning Resources

Community